CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   ANSYS Meshing & Geometry (http://www.cfd-online.com/Forums/ansys-meshing/)
-   -   [ICEM] 3D turbine blade modeling (http://www.cfd-online.com/Forums/ansys-meshing/72858-3d-turbine-blade-modeling.html)

ShaneStan February 19, 2010 17:10

3D turbine blade modeling
 
Hi guys, I'm quite new to Fluent, ICEM and Gambit. I have to model a rotating turbine blade in 3d. I'v only done 2d analysis before now so im having trouble with this! I also have no idea how to rotate the blade! i have the geometry made. ICEM is supposed to be more user friendly for 3d i hear, so i am favouring this i think?! any suggestions or help would be much appreciated!! thanks

PSYMN February 20, 2010 14:41

If you can find it, there is a "stator" tutorial for meshing in ICEM CFD. That would be a good starting point. I would send it to you, but I haven't seen it for a while myself. It may not have been included with the most recent versions, but I guess I may have just missed it.

If you are really new, you may want to skip all the Hexa meshing and go for a Tetra Prism mesh until you get the solver details worked out.

Once you get into hexa meshing, you will need a different topology depending on if you have rounded or sharp leading and/or trailing edges, tip gaps, squealers, etc. It is not the sort of thing I can explain in text. :eek:

One thing for sure, you won't need to mesh the entire model, just periodic segments of the rotating components.

ShaneStan February 23, 2010 10:27

thanks for your reply psymn,
I found that tutorial i think you were talking about, but its for cfx?! is cfx much different then icem of gambit? i dont know much about it. :(

Zaqie February 23, 2010 23:47

3d turbine with periodicity meshing
 
I am doing a similar work. Ocean current turbine. I have done tetra-prism meshing with a periodic boundary.

Now I am trying to do hexameshing.

1. Is hexameshing possible with periodic boundary?
2. I have seen an airplane hexameshing in ICEM CFD tutorial. can hexameshing be done similary with periodic boundary. I have two periodic boundary with three sectors(120 degrees each)?

Please help.

Regards

Zaqie

ShaneStan February 24, 2010 09:56

i found that tutorial simon-its for fluent 6.3!! its clearing things up for me-thanks again for your reply!!
shane

PSYMN February 24, 2010 11:14

I guess there are lots of “Stator” tutorials out there... There was an ICEM CFD Hexa one also. It covered Periodic blocking (yes you can make ICEM CFD Hexa Periodic).

To make an ICEM CFD Hexa mesh Periodic, you first setup the periodicity in the Global settings under the mesh tab...

Once in the Hexa Tab, you can make Vertices periodic. Once you make the 4 verts around a face periodic, the face will become periodic and any further splits will already be periodic.

The 12.1 tutorials are now online but do not yet include the Stator. I did post the location for the 11.0 tutorials somewhere on CFD-Online. It may be in there.

Zaqie February 25, 2010 15:44

tutorial not found
 
Simon,

The stator tutorial cannot be found in ICEM CFD 11. Is there a way I can get it?

Thanks

Zaqie Reza

ShaneStan February 27, 2010 15:27

icem 10 tut
 
hey zaqie,
i came across the icem 1o set of tutorials-i know they'r not what your looking for but they might be of some use!!! if you want it send me ur email because this wont let me upload it for some reason!!
shane

Zaqie February 27, 2010 16:21

my email
 
my email

zaqie.reza@gmail.com

thanks Simon

ShaneStan February 28, 2010 10:24

General Questions for sliding mesh creation on/around a turbine blade [ICEM]
 
What is the appropriate surrounding geometry for a problem of this kind?

How should the blocks be split?

What are the appropriate associations to do?

Is an Ogrid suitable within this application?

What boundary conditions should be implemented?


Keep an eye on this Thread if your in need of answers to these questions too...
Shane

PSYMN March 1, 2010 16:19

Blocking strategy
 
5 Attachment(s)
Hello all,

ShaneStan sent me his files and I just took a quick look… He was having trouble blocking this airfoil and just needed help getting a good topology. He was particularly having trouble with his OGrid… His airfoil has lots of camber and a fair amount of twist and it has a sharp trailing edge (which makes all the difference for blocking).

Keep in mind that there are two primary uses of OGrid. First, is for capturing flow topology. You usually do these sorts of OGrid operations early in the blocking and they really help you get ahead and capture the topology. The second type is for boundary layer flow or simply to resolve problems where hexas are being opened out to 180 degress along a curved corner. You usually look after that at the end of a blocking after pretty much everything else is in place. I will just talk about the first type for this one.

Specific to Shane’s model, he had the airfoil floating in the middle of a large cube, which is odd. Normally, I would expect the root to be against a symmetry plane or something like that. Perhaps he had some good reason.

Shane has the input flow moving towards negative Z… But these sorts of high lift configurations tend to deflect the flow downward, so I decided to start with a Quarter OGrid (aka YGrid) to bend the flow along with the airfoil. This is done with the OGrid tool and faces on 4 of the 6 sides. This isn’t necessary, but it does help to concentrate the elements where they are most needed. Later, I could perhaps slope the downward edges so they go back at 45 degrees or something depending on how the wake actually looks.

As I mentioned, his model has a sharp trailing edge. You should use “collapse block” to capture this. In this case, it will be best to first create the boundary layer around the airfoil, then collapse out the middle trailing block…

To start, I would split ahead of the airfoil, and above and below it. This creates a region around my airfoil. Note I have not yet split anything across X; for simplicity, I am doing all my work in 2.5 D and would fit it to the 3D model later. (This prevents me from needing to deal with 7 noded blocks after the collapse) I then put a CGrid in this airfoil boundary zone… I have an image showing the selected blocks and faces for this Cgrid.

Finally, you can “collapse” block to pinch off the trailing edge.

After this, Shane still has a fair bit of work to make his other splits and associations and to setup sizes, etc. He basically just needs to fit the “triangle” to his airfoil. He should do that with associations and then use the move vertex command to tweak things a bit. But the rest is pretty straight forward. Don’t forget to convert from the pre-mesh to the unstructured or structured multi-block format before outputting.

As for bocos, he already has put his inlet and outlet into named parts, so he will just need to go to the output menu, select his solver and create a boco file. Many of the formats, like Fluent, have a lot of supported bocos. But even if you don’t want to use these, at least open the boco menu and hit apply on the popup. Then make sure to save the project before outputting because the output executables run on the saved versions of the files...

Best regards,

Simon

PSYMN March 1, 2010 16:21

Last Pic...
 
1 Attachment(s)
CFD-Online only lets me post 5 pics, and I had 6;)

Here is the last one after the collapse of the trailing edge block.

Now Shane needs to split the other direction for the airfoil root and tip and associate the edges as required.

He may want to use "align" to straighten out the blocking on the sides of the box with the blocking at the ends of the airfoils.

ShaneStan March 2, 2010 09:20

Just for when i'm doing the associations-what blocks am I associating to the airfoil's? several different triangular blocks pass theought the airfoil and its hard to see which are the correct ones?:confused:

Zaqie March 2, 2010 17:14

turbine blocking
 
Hi PSYMN

I have attached couple of pictures of my turbine and the domain.

I totally understand the idea of Ogrid and blocking, but am finding hard to do the same to my turbine.

I created a 3d bounding box and split at the turbine area so that I can associate surfaces of the turbine with the block surface. And then tried to create the Ogrid(just like in the wing body example). But I am not able to premesh.

Please look the attachments. Am i doing the correct thing or do I have to take an approach like the one you suggested to Shane?

Regards

Zaqie Reza

Zaqie March 2, 2010 17:52

turbine blocking
 
2 Attachment(s)
sorry i forgot the attachments

PSYMN March 2, 2010 20:42

Similar... But Periodic.
 
Zaqie,

If your model has a sharp trailing edge, then you should do similar to what I showed Shane, except that your model is periodic, so it requires a different treatment... Perhaps just pinch the block down for a distance and not all the way to the periodic boundary...

You also don't need to do the 90 degree turn of the first Ogrid that I did, your splits can just cut across from one periodic face to the next.

Your pictures don't show what you have tried, so I can't really comment beyond that.

Simon

PSYMN March 2, 2010 20:54

Final steps... 1 of ...
 
4 Attachment(s)
Shane,

Lets see if we can finish this off. It will take a few posts to get thru all the images I just snapped... But there are not really many steps and it is not difficult...

I saw your previous blocking, so I am assuming that you know about association, etc. and are really just asking for the bigger picture steps… (Good, because that is all I have time for today ;^)

First, rotating the model and looking from the front, I see the wing floating strangely… I need to create two splits to capture the root and tip… I didn’t create these two carefully, just with screen select, but if I were scripting this, I would have used “prescribed point”. I also showed these from a different vantage point.

Nest reduce the index control so you are just looking at the little wedge that will represent the wing (I just had to raise 04 so it was from 1 to 1). This is what needs to be associated to the little wing.



To be continued...

PSYMN March 2, 2010 20:57

Final steps... 2 of ...
 
5 Attachment(s)
Now it is time to associating the trailing edge…

Next, I would handle the triangles on the sides… After moving things into place, I am reminded that the camber will affect projection. I could use “edge splits” to steer things, but since Shane has already got these curves along the airfoil (root to tip), I will split the airfoil and hang on to those. (I reduced my "i" index, so I was only splitting the airfoil block).

Then I needed to Delete the internal blocks (blocks inside the airfoil).


To be continued...

PSYMN March 2, 2010 21:01

Final steps... 3 of ...
 
4 Attachment(s)
I set sizes on the airfoil (just put 10 nodes in each section), matched edges, etc.

But now that I have fit the triangle to the twisted airfoil, things look pretty twisted… (in the second picture) to solve this, I use the Align command.



Start by reducing the index range so you see from the airfoil tip to the FF boundary (align only works on the active index and we don’t want it messing up the root side of the airfoil). Align in the X direction using the airfoil tip for a reference… I have attached a full screen shot so you can see the GUI and the align tool in the bottom left and the index control in the bottom right…


To be continued...

PSYMN March 2, 2010 21:06

Final steps... 5 of 5
 
5 Attachment(s)
Then go to the ff boundary and straighten things out there (don’t move the little triangle). You can see the image with the mesh straightened out and the surface blocking straightened out...



Then align back again so that your straightened out surface blocking can straighten out the volume plane around the wing tip. Note it is easier to move things around on the surface which is why we did it this way.

Then do the same for the Root side to its far field and back again.

Then I spent a few minutes massaging back and forth, setting edge distributions, etc. and tada! Done.



I hit clear on the replay script and there were about 130 lines left total, including all my edge parameter settings, etc.

Well, I am done. You could still tweak it from here to improve quality, distribution, etc.


All times are GMT -4. The time now is 01:15.